sync to latest upstream source code

actions.hpp

@@ -311,8 +311,6 @@ json action_sampling_init(app_t &app, json &body)
     sparams.penalty_freq = body["penalty_freq"];
   if (body.contains("penalty_present"))
     sparams.penalty_present = body["penalty_present"];
-  if (body.contains("penalize_nl"))
-    sparams.penalize_nl = body["penalize_nl"];
   if (body.contains("dynatemp_range"))
     sparams.dynatemp_range = body["dynatemp_range"];
   if (body.contains("dynatemp_exponent"))
@@ -580,7 +578,7 @@ json action_embeddings(app_t &app, json &body)
       return json{{"error", "failed to get embeddings"}};
     }
   }
-  common_embd_normalize(embd, out, n_embd);
+  common_embd_normalize(embd, out, n_embd, 2);
   return json{
       {"success", true},
       {"embeddings", embeddings},

package.json

@@ -1,6 +1,6 @@
 {
   "name": "@wllama/wllama",
-  "version": "2.1.0",
+  "version": "2.1.1",
   "description": "WebAssembly binding for llama.cpp - Enabling on-browser LLM inference",
   "main": "index.js",
   "type": "module",

src/wasm-from-cdn.ts

@@ -2,8 +2,8 @@
 // Do not edit this file directly
 
 const WasmFromCDN = {
-  'single-thread/wllama.wasm': 'https://cdn.jsdelivr.net/npm/@wllama/[email protected].0/src/single-thread/wllama.wasm',
-  'multi-thread/wllama.wasm': 'https://cdn.jsdelivr.net/npm/@wllama/[email protected].0/src/multi-thread/wllama.wasm',
+  'single-thread/wllama.wasm': 'https://cdn.jsdelivr.net/npm/@wllama/[email protected].1/src/single-thread/wllama.wasm',
+  'multi-thread/wllama.wasm': 'https://cdn.jsdelivr.net/npm/@wllama/[email protected].1/src/multi-thread/wllama.wasm',
 };
 
 export default WasmFromCDN;

src/wllama.ts

@@ -108,7 +108,6 @@ export interface SamplingConfig {
   penalty_repeat?: number;
   penalty_freq?: number;
   penalty_present?: number;
-  penalize_nl?: boolean;
   dynatemp_range?: number;
   dynatemp_exponent?: number;
   grammar?: string;